A new skeleton of lanthasaurus hardestii, a

Transcription

A new skeleton of lanthasaurus hardestii, a
834
A new skeleton of lanthasaurus hardestii, a primitive edaphosaur
(Synapsida: Pelycosauria) from the Upper Pennsylvanian of Kansas
S. P.
MODESTO AND R. R. REISZ
Department ojZoology. ErilUiale Campus. University ojToron/o. Mississauga, On/., Canada LSL lC6
Received September 11, 1989
Revision accepted February 7, 1990
A new specimen from the Upper Pennsylvaman of Garnett, Kansas, is referable to the edaphosaur Ianthasaurus hardestii.It
is the second articulated skeleton known of this species, aud possesses prevlOnsly undescribed midline elements of the skull roof.
This specimen featnres a new autapomorphy, an elongate cross-barred dorsal process on the axial ueural spine. The presence of
veutral webbing and multiple tubercles on several basal lateral protuberances of the presacral neural spines in this skeleton may
represent a sexual dimorphism of Icmthasauru5. Two synapomorphies of sphenacodonts and edaphosaurs, the presenee of a
laterallapper on (he frontal and exclusion of the reduced quadrar.ojugal from the ventral margin of the temporal bar, are confirmed
in this specimen. The presence of these traits in Iantha.murus supports the hypothesis that edaphosaurs and sphenacodonts form a
clade more derived thau other pelycosaur groups. New information provided by this specimen indicates that edaphosaurs can be
recognized only by the morphology of their distinctive presaeral neural spines. Small, problematical edaphosaur species assigned
to the genus Edaphosaurus may be reinterpreted as iusectivores, closely related to the Garnett edaphosaur and distinguishable
from the large, bulky herbivore Edaphosaurus.
Un nouveau specimen du site Garnett (Pennsylvanien superieur) est assigne al'ectaphosaure lanthasaurus hardestii. II s'agit
du second squelette articuIe connu de cet edaphosaure. Ce squelette possede des elemems medians du toit cranien non preserves
dans l'autre specimen de ce genre. Ce specimeu demontre la presence d'une nouvelle autapomorphie : I'epine neurale axiale
possede un processus transverse dorsal allonge. La presence d'une palmure ventrale ct de tubercules multiples sur plusieurs
protuberances lateraIes proximales des vertebres presacrees indiquent peut-etre un dimorphisme sexuel. L'existence de deux
synapomorphies des spMnacodontes et des edaphosaures, la presence d'un processus hiteral du frontal et l'exclusion du
quadratojugal du bord ventral de l'arc temporal, est confirmee par ce specimen. L'edaphosaure de Garnett posse-de des caractees
qui supportent l'hypothese selon laqulle les Cdaphosaures et les sphenacodontes forment un clade plus derive que les autres
pelycosaures. Ce nou veau specimen indique que 1'Edaphosauridae peut erre definie seulemem par la morphologie de leurs epines
neurales presacrees. Les petites especes d'edaphosaures d'affinites incertaines qui avaient ete assignees au genre EdaphosauJ"US
peuvent etre reinterpretees comme etant des insectivores etroitement apparentes al'edaphosaure de Garnett et dis tincts du grand
herbivore Edaphosaurus.
Can. J. Earth Sci. 27, 834-844 (1990)
Introduction
(1986) suggested that this edaphosaur fed primarily upon
The pelycosaur family Edaphosauridae, represented by the
genera Edaphosaurus and lanthasaurus, spans the Upper
Pennsylvanian and the Lower Permian strata of Europe and
North America.
Edaphosaurus is one of the most widespread and successful
of Pennian reptiles. Eight species are currently recognized
(Reisz 1986). Dental morphology indicates that Edaphosaurus
probably fed upon plant material. The majority of differences
between Edaphosaurus and lanthasaurus are directly attribut­
able to this adaptation to herbivory: large body size, barrel­
shaped trunk, rel~tively small skull, enlarged temporal fenes­
trae, slightly bulbous marginal dentition, and massive crushing
tooth plates on the palate and the medial surface of the
mandible. Edaphosaurus is adequately represented by five
North American species, four of these by cranial as well as
postcranial material. In the absence of adequate specimens, the
validity of two small European and a small North American
species has been questioned by Reisz (1986).
Ianthasaurus, the smaller, earlier, and morphologically more
primitive member of the Edaphosauridae, is known from a
single species from the Upper Pennsylvanian of Kansas.
lanthasaurus is distinguished from later edaphosaurs by its
small body and dorsal-sail size, slim trunk, relatively large
skull, small temporal fenestrae, long and low maxillae with
sharp, recurved teeth, caninifoffil teeth, and no tooth plates
(Reisz and Bennan 1986). On the basis of the morphology of its
marginal and inferred palatal dentition, Reisz and Berman
soft-bodied insects. The small, elongate body, large head, and
small temporal openings support this hypothesis. Therefore,
lanthasaurus differs from Edaphosaurus essentially in its
adaptation to insectivory.
The description of lanthasaurus hardestii (Reisz and Bennan
1986) was based solely on material from Garnett, Kansas. The
Garnett locality has produced an abundance of vertebrate, inver­
tebrate, and plant remains, including taxa of great phylogenetic
signiticance such as the oldest known diapsid Petrolacosaurus
kansensis (Reisz 1977, 1981), the earliest sphenacodont Hap­
todus garnettensis (Currie 1977), and other unique, endemic
amniote species (Reisz et at. 1982; Dilkes 1987). The speci­
mens available to Reisz and Berman (1986) depicted a very
primitive eupelycosaur with a single autapomorphy. the pres­
ence of at least 29 presacral vertebrae. The holotype also
characterized lanthasaurus as bearing a maximum of five pairs
of lateral tubercles on each neural spine, with crossbars
concentrated on the anterior two thirds of the sail. However, the
palate, braincase, occiput, most of the skull table, and much of
the pectoral girdle were absent from the known specimens or, in
the case of the limb elements, too poorly preserved for
descriptive purposes.
The description of the Garnett edaphosaur appeared soon
after recent reevaluations of pelycosaur interrelationships.
Brinkman and Ebelth (1983) showed that the Edaphosauridae
and the Sphenacodoncidae form a clade defined by several
synapomorphies. Reisz and Berman (I986) demonstrated that
MODESTO AND REISZ
Ianthasaurus possessed three of the 10 synapomorphies of the
edaphosaur-sphenacodont clade, and they inferred the presence
of two more.
The subject of this paper is a well-preserved and partially
articulated skeleton of I anthasaurus collected by G. MacDonald
and M. Heaton from Garnett. The significance of this specimen
lies in the presence of most of the skUll-roof elements. Closely
associated with the skull is an assemblage of 23 presacral verte­
brae, including the axis, and a number of other disarticulated
elements. The new information provided by this specimen
allows for a revised diagnosis of the genus and more complete
reconstructions of the skull and sail than those offered by Reisz
and Bennan (1986).
The nearly complete skull roof also permits more thorough
comparisons with later members of the family. Ianlhasaurus is
compared with EdaphosauTus novomexicanus and Edaphos­
aurus boanerges not only because these species are the oldest
and most primitive members of the genus Edaphosaurus, but
also because specimens of these species are available for direct
comparisons.
The new material prompts a discussion of edaphosaur
phylogeny and the probable identity of the small, poorly known
edaphosaur species.
Systematic paleontology
SUBCLASS Synapsida
ORDER Pelycosauria
FAMILY Edaphosauridae Cope, 1882
Revised diagnosis
Small to large pelycosaurs with greatly elongated presacral
neural spines, which are subcircular in cross section except for a
short, laterally compressed proximal portion. Neural spines of
the cervical region lean anteriorly, whereas those of the lumbar
region curve posteriorly. Most presacral neural spines possess
lateral tubercles of regular distribution and development; basal
lateral tubercles are always paired.
GENUS Ianthasaurus Reisz and Berman, 1986
lanthasaurus haTdestii Reisz and Bennan, 1986
Revised diagnosis
A small edaphosaur characterized by an elongate, Cross­
baned dorsal process on the axis and at least 29 presacral
vertebrae. IanthasauTus differs from other edaphosaurs in the
following features: skull long, equal to eight dorsal centra;
frontal long and broad anteriorly, with lateral lappet located just
posterior to midpoint; parietal with greatest breadth at level of
pineal foramen; maxilla elongate with 27-29 sharply pointed,
slightly recurved teeth; caniniform teeth present; no develop­
ment of enlarged tooth plates on the lower jaw; postaxial cervical
vertebrae slightly longer than succeeding presacral vertebrae.
neural arches of presacral vertebrae laterally excavated and
bearing short transverse processes; maximum of five lateral
tubercles on each side of neural spine, with some proximal
tubercles supported ventrally by slight webbing; no tubercles on
neural spines of the posterior region of presacral column; trunk
ribs strongly curved proximally and bearing well-developed
tubercula; ilium with well-developed, blade-like posterior
process.
Holotype
KUVP 69035 (University of Kansas Museum of Natural
History) consists of a nearly complete series of articulated
835
presacral vertebrae, several disarticulated dorsal ribs, and
several disarticulated skull-roof elements of the right side of the
skull. A humerus is also present, as is the posterior two thirds of
the lower right jaw.
Attributed specimen.s
ROM 29942 (Royal Ontario Museum) includes several
scattered cranial and postcranial bones (also present is an
amphibian ilium and several poorly preserved and unidentifi­
able limb elements). ROM 29941 consists of three isolated
vertebrae assigned to a single catalogue number for convenience
only. ROM 29940 consists of a fust sacral rib, seven caudal
ribs, and a partial dorsal neural arch with a fragmented but
complete neural spine. ROM 37751, the subject of this study,
consists of a semiarticulated skull roof, left angular, both ilia,
left pubis, left ischium, 23 presacral vertebrae, several caudal
vertebrae, and dorsal and caudal ribs on three closely associated
blocks.
Horizon an.d locality
Rock Lake Shale Member of the Stanton Fonnation, Lansing
Group, Missouriau Series, Upper Peuusylvauian, NW 1/4, sec.
5, tp. 19E, Putnam Township, Anderson County, Kansas.
Description
Although not perfectly articulated, ROM 37751 (Figs. 1-3)
undoubtedly represents the remains of a single animal; all of the
elements lie on the same bedding plane, there is no duplication
of cranial and appendicular elements, and the preservation of all
elements is identical.
The skeleton is that of an immature animal: none of the neural
arches are fused to their centra, the preserved pelvic girdle
elements are not coossified, and the cranial elements are of
comparable size to those present in the holotype, which has also
been shown to be immature (Reisz and Bennan 1986).
The skull roof is nearly completely preserved except for the
tooth-bearing elements. Problems associated with the removal
of the specimen from the quarry resulted in its separation into
three small blocks. Consequently, some bone was lost from the
skull roof, primarily from the right prefrontal, left postfrontal,
and both frontals. Otherwise, most cranial elements are undam­
aged and the paired dorsal elements have retained their midline
sutural contacts. We achieved precise realignment of the
separated skull table by measuring the undisturbed distances
between the frontal-postfrontal suture at the orbital border and
two different points on the midline along the right frontal and
repositioning the two blocks in such a way that the complemen­
tary measures were restored on the left side.
A revised reconstruction of the skull roof in lateral aspect is
shown in Fig. 4a. This reconstruction differs from that offered
by Reisz and Berman (1986) in three respects: (1) the orbit is
slightly larger than in the original description because of the
undetected absence of a posterior process on the holotypic
prefrontal; (2) the jugal is more slender; and (3) the temporal
fenestra is ~maller. These corrections in the reconstruction of
the skull of IIanthasaUTUS in lateral aspect result in a much more
gracile skull than that suggested by Reisz and Bennan (1986).
The presence of most of the skull-roof bones permits a
reconstruction of the skull in dorsal perspective (Fig. 4b). The
exact width of the skull roof, however, remains conjectural
because palatal elements are unknown. A partial reconstruction
of the skeleton (Fig. 5) provides new information on the nature
836
CAN J EARTH SCI. VOL. 27, 1990
FIG. 1 lanthasaurus hardestii, ROM 37751. Partial skull roof in ventral view, angular, partial vertebral column, and pelvic girdle elements.
Scale is in centimetres.
MODESTO AND REISZ
837
FIG. 2. lantlwsaurus hardestii, ROM 37751, referred specimen. Partial skull roof exposed in ventral view, angular, partial vertebral column,
and pelvic girdle elements. Abbreviations: ang, angnlar; ax, axis; f, frontal; iI, ilium; is, ischium; j, jugal; I, lacrimal; n, nasal; p, parietal;
pf, postfrontal; po, postorbital; pp, postparietal; prf, prefrontal; pu, pubis; qj, quadratojugal; r, dorsal rib; sq, squamosal; 3-21,24-26, presacral
vertebrae. Scale = I crn.
838
CAN .l EARTH SCT VOL 27, 1990
~
~
.•.. ,.
'f~'~ .' "
..•
.
':."
.. ('!Pf..
FIG. 3. lanthasaurus hardestii, ROM 37751, referred specimen. Partial skull roof exposed in dorsal view, angUlar, partial vertebral column,
and pelvil: girl1le elements. See Fig. 2 for abbreviations. Scale = 1 em.
839
MODESTO AND REISZ
of the crossbars and Illustrates the relationship between head
and sail.
The following descriptions reter to the specimens seen in
Fig~.
2 and 3. For tho~e elemem~ that are not de~crihed in demi1
below refer to the initial description (Reisz and Rennan 1986).
The well-preserved nasals are large, long. trapezoidal sheets
of bone that dominate the snout (Pig. 2). The nasal is almost as
long 31; the frontal at the midline. The bone is slightly convex in
transverse sedilln, het;omi ng progressl vel y flatter anteriorly. A
small oval foramen on the ventral sutiace, its long axis aligned
anteroposteriorly, probably represents the egress of the orbito­
nasal vein. Laterally, the nasal articulates with the prefrontal
and lacrimal along a faintly sigmoidal suture. A small anterior
process of the nasal abuts its fellow and fonns half of a small
wedge between the dorsal processes of the premaxillae. To what
extent the premaxilla overlapped this process remains unknown
because the overlying ischium obscures this part of the nasals in
dorsal view (Fig. 3). A slight concavity on the anterolateral
border of the nasal probably represents the posterodorsal margin
of the external naris. The nasal-frontal contact, obscured by the
lacrimal in dorsal view (Fig. 3) and only partially visible in
ventral view (Fig. 2), is a strongly interfingering suture. The
nasal underlies a loug, tongue-like process of the frontal
(Fig. 3).
Both lacrimals are present but poorly preserved. In each the
lacrimal duct has been exposed through loss of the lateral
surface.
The fmntals are well preserved, except for the loss of bone
explained above (Figs. 2, 3). The strongly developed lateral
lappet lies approximately two thirds back from the nasals and ou
its dorsal surface features very shallow but perceptible trans­
verse scalloping. The free lateral border of the lappet is
approximately one quarter of the length of the frontal at the
midline. An interfingering suture between frontal and parietal is
evident in dorsal view (Fig. 3), but ventrally this interdigitation
is not pronounced (Fig. 2). Medially, the frontal receives a
small, tongue-like anterior process of the parietal. Lateral to
this, the frontal bears a long posterior process that is received by
the parietal and contributes significantly to the central parasagit­
tal ridge of that element. Articulation with the postfrontal is
long and concave. The anterior process of the frontal tapers to a
truncated edge that contacts the nasal. This process is approxi­
mately twice as long as the posterior process and bears
distinctive striations that suggest that growth at this stage was
occurring primarily to the front.
Both prefrontals differ from the description of the holotype
(Reisz and Berman 1986) in possessing a well-developed pos­
terodorsal process that bears shallow scalloping on its dorsal
surface. Accompanying postdepositional crushing, separation
of the main bodies of the prefrontals in ROM 37751 from their
posterodorsal processes occurred. These processes retain their
respective sutures with the frontals, and presumably this may
also be the case with KUVP 69035. Therefore, the prefrontal
contributes more to both the dorsal skull roof and the orbital
margin than Reisz and Bennan (1986) have suggested.
Both postfrontals are present and do not differ significantly
( b)
"-'''':''.
FIG. 4. lanthasaurus hardestii. Composite reconstruction of skull in
lateral (a) and dorsal (b) views, based mainly on ROM 37751. Scale =
1 cm.
The postorbitals assume the same outline as that in the
holotype. The left element, the best preserved of the pair, shows
three characteristics not previously observed. Weak scalloping
is present on the lateral surface of the raised orbital margin, and
the posterior process has a distinct ridge along the edge
articulating with the parietal. The most surprising feature is the
small size of the posterior process. The sutural pattern of this
process with the squamosal and parietal indicates that the
postorbital did not contact the supratemporal.
The parietals remain fully articulated with both frontals and
postfrontals. The parietal is approximately two thirds of the
length of the frontal. The subcircular pineal opening lies just
posterior to the midpoint of the interparietal suture The pineal
foramen features a slight dorsal lip, which contributes posteri­
orly to a broadly rounded ridge fanned by both parietals at the
midline. This ridge may have anchored an anterior extension of
connective tissue from the neural spine of the axis. The lateral
edge of the parietal, overlapped anteriorly by the postfrontal, is
moderately convex in dorsal view, with the parietal reaching its
greatest width at the level of the pineal opening. The posterolat­
eral wing of the parietal is approximately one fifth of the total
anteroposterior length of the bone and bears a relatively deep,
narrow groove for the supratemporal. Lateral to this groove, the
parietal has a small lateral flange that overlies the postorbital.
The posterior margin of the parietal is strongly concave and
bears a shallow shelf for receiving the tabular. A parasagittal
ridge extends posteriorly from the point where the frontal and
postfrontal meet, curves laterally at the level of the pineal
foramen, and becomes progressively rounded until it dissipates
at the level of posterior edge of the pineal opening.
The single median postparietal, unusually thin, contacts the
parietals anteriorly but is impetiectly defined posteriorly and
from those seen in KUVP 69035 and ROM 29942, except that
laterally.
the scalloping observed by Reisz and Berman (1986) is more
pronounced and occupies the lateral two thirds of its dorsal
surface. The large anterior expansion of the postfrontal inter­
preted for ROM 29942 by Reisz and Berman (1986) is actually
the posterior portion of that element.
Both jugals are present, but only the left element is well
preserved and retains most of its sutural contacts. Neither differs
from that described by Reisz and Berman (1986) except in
having the anterior edge of the dorsal ramus slightly thickened
and continuous with the orbital rim on the postorbital.
840
CAN. J. EARTH SCI. VOL. 27, 1990
FIG. 5. lanthasaurus hardestii. Composite reconstruction of skull
and dorsal sail in left lateral view, based mainly on ROM 37751. Scale
= 1 cm.
The left squamosal is preserved in articulation with the
postorbital, wheras the suture with the jugal has been displaced
slightly. This squamosal differs little from the squamosal of the
holotype (Reisz and Berman 1986), except that the posterior
flange of hone that supports the supratemporal is preserved in
ROM 37751. The flange arises from the posterionnost point of
the squamosal-parietal contact and ends directly opposite the
ventral margin of the subremporal ramus.
The left quadratojugaI is preserved in articulation with the
squamosaL Although the quadratojugal is obscured to some
extent by a neural spine medially and a caudal neural arch
laterally, the trapeLoidal outline of the quadratojugal can be
discerned. Ventrally the quadratojugal is relatively thick where
it probably abutted against the condylar process of the quadrate.
Dorsally the quadratojugal becomes progressively thinner,
tenninating with a rounded dorsal bordcr conflucnt with the
medial surface of the squamosal. The element is well removed
from the lower temporal bar.
Only the left angular remains of the lower jaw; it appears to
be essentially complete. It is a long, moderately deep element
whose ventral e<1ge i~ ~(raight for much of its length but curyes
gently upwards about one fifth from its posterior end. The
medial surface of the ventral keel bears weak striations that
suggest growth was occurring downwards, deepening the keel.
Notably absent is the area associated with the articular. Laterally,
the anterior half of the angular is divided into dorsal and ventral
portions by a prominent ridge (Fig. 3). The dentary was probably
sutured to the anterior region dorsal to this ridge.
Approximately 23 presacral and eight caudal vertebrae are
preserved. Centra are absent except for those of two distal
caudal vertebrae. Most of the vertebral arches and their spines
suffered some crushing and distortion. The neural arches of the
second cervical through to the twenty-sixth vertebra lie closely
associated and are posterior to the skull. They were laid down in
what appears to be a semiarticulated state, preserved as though
the sail membrane that once held the vertebrae together had been
folded at three points. The first series of vertebrae (Fig. 3)
consists of the axis through to the seventh vertebra, at which the
first fold occurs. The second series overlies the first (Fig. 3). It
consists of the eighth vertebra through to the thirteenth, after
which the second fold occurs. Passed back directly over the
second series, the third series (Fig. 2) comprises the fourteenth
vertebra through to the twenty-first. The orientation of the last
three (Fig. 3) is the reverse of that of the other vertebrae. The
preservation of both the dorsal surface of the skull and the tirst
vertebral series on the same level, coupled with the close
proximity of the axis to the posterior end of the skull, supports
the hypothesis that the skull and axial neural spine were
connected by an anterior extension of the sail. Prior to
entombment and after the decomposition of the soft tissues, the
sknll of ROM 37751 apparently moved away from the axis, and
the neural spines were rearranged as indicated above.
The neural arches and spines are similar in most respects to
those desclibed by Reisz and Bennan (1986). The dorsal sail of
ROM 37751 differs from the holotypic sail in that the 12 neural
spines posterior to the axis are approximately 20-25% taller.
The remaining neural spines differ little in height from those of
the holotype. This difference may result from a period of
differential growth in the sail or possibly sexual dimorphism.
Lateral excavations of the neural arches are preserved in those
vertebrae that did not suffer extensive distortion from being
crushed against underlying vertebrae.
MODESTO AND REISZ
The axial neural arch (Fig. 3) is readily idemified by dle
presence of small anterior zygapophyses that project only
slightly from the lateral surface of the arch and the anteropos­
teriorly expanded spine that presumably projected over the
posterior end of the atlas neural arch and served for attachment
of nuchal ligaments. The posterior horder of the antcropos­
teriorly expanded pordon of the spine is relatively thick, whereas
the anterior margin is quite thin. Directly ahove the posterior
zygapophyses, a tall neural spine similar to those home by the
postaxial neural arches extends dorsally. This spiue, although
broken at the tip, appears typically edaphosaurian in nature,
being subcircular in cross section and curving forward at the
base and slightly backwards distally. A small lateral tubercle
can be seen on the right side (Fig. 3), approximately at the same
level as basal tubercles of the postaxial neural spines.
The postaxial spines show the typical edaphosaurian arrange­
ment of laterally projecting crossbars that form longitudinal
rows along the dorsal sail. The postaxial series of ROM 37751
conforms closely to that described by Reisz and Bemlan (1986)
for the holotype. Lateral tubercles are rare on the posterior
dorsals and totally absent on the lumbar neural spines. The
anterior dorsals have no more than five pairs of crossbars, with
mid-dorsals exhibiting crossbars only on the proximal two
thirds of the spine. As in other specimens of this genus, the
tubercles project laterally and slightly dorsally, with the largest
lateral tubercles located proximally on the spine. Some of the
largest tubercles feature small, web-like ventral sheets, which
may have served to strengthen the tubercle. The webbing is
considerably thinner than dle tubercle itself and tapers to a
rounded edge. This feature was not observed in the holotype or
ROM 29942, possibly because ofthe manner of exposure of the
spines or because the feature may have been so poorly de­
veloped as to elude observation. In addition to the webbing,
accessory tubercles arc borne by several basal tubercles. The
seventh vertebra is interesting: it features both multiple tubercles
and ventral webbing. As in the holotype and other specimens,
the tubercles distal to the basal pairs are less prominent and
accordingly may feature weaker webbing. The pairing of lateral
tubercles becomes more irregular distal to the basal pair,
occasionally assuming a staggered appearance.
Several ribs from the dorsal, lumbar, and caudal regions are
preserved but do not add new information to the original
description of the genus (Reisz and Berman 1986).
A pubis) an ischium, and both ilia are preserved. The ilia
(Fig. 3) appear well preserved with the exception ofsome loss of
bone along the anterior edge; both exhibit the blade-like posterior
process observed by Reisz and Berman (1986). Neither element
possesses a well-defined ventral border because ofthe immatu­
rity of the animal. The left pubis underlies two posterior dorsal
neural spines. An overlapping fold along the posterior edge of
this bone probably represents the location of the obturator
foramen. The contribution to the acetabulum may be represent­
ed by a slightly thickened, smooth region of bone at the margin
next to the obturator foramen. The remainder'of the element is
otherwise nondescript. The ischium, identified by its thick,
concave dorsal margin, underlies the anterior portion of the
skull. Because of crushing, marginal distortion, and loss of
bone, its assignment to the left side is tenuous.
Discussion
In 1940 Romer and Price suggested that edaphosaurs and the
younger caseids were two distinct herbivorous llneages of the
841
suborder Edaphosauria. Four decades larer Kemp (1982) and
Reisz (1980), employing cladistic analysis, hypothesized that
edaphosaurs were more closely rclatcd to the carmvorouS
ophiacodontids and sphenacodontids. Following a more rigor­
ous analysis, Brinkman and Ebenh (1983) hypothesized that
edaphosaurs and sphenacodonts formed a clade defined by
several synapomorphies. The latter theory implies that herbiv­
ory had evolved within the Edaphosauridae, suggesting thac [he
group must have passed through a carnivorous stage, possibly as
insectivores. This was confirmed by the discovery of the
insectivorous edaphosaur lanthasaurus (Reisz and Berman
1986). The osteology of Ianthasaurus also provides direct
evidence for the hypothesis of close relationships between
edaphosaurs and sphenacodonts (Brinkman and Eberth 1983).
The initial description (Reisz and Berman 1986) showed that
Ianthasaurus possesses the following derived features of
edaphosaurs and sphenacodonts: (1) a preanicular twisted
posteriorly as to underlie the pterygoideus process of the
articular, (2) a pterygoideus process formed by the articular
alone, and (3) excavation of the lateral surfaces of the neural
arches.
Reisz and Berman (1986) inferred the presence of two
additional synapomorphies shared with sphenacodonts: a small
quadratojugal excluded from the temporal bar, and a frontal
lateral lappet. A small quadratojugal located medial to the
squamosal in ROM 37751 confirms the first. The quadratojugal
of Ianlhasaurus is similar to that of Haptodus in its configura­
tion and small size. Reisz and Berman (1986) deduced the
presence of the frontal lateral lappet from the wide gap between
the prefrontal and postfrontal elements in their tentative
reconstruction of the skull. The contribution of the frontal to the
orbital margin in ROM 37751 is approximately that postulated
by Reisz and Berman (1986). This is so despite the observation
that the absent posterodorsal process of the holotypic prefrontal
was not detected by Reisz and Berman (1986). The frontal
lappet of lanlhasaurus assumes the most primitive state
observed in edaphosaurs, being broad anteroposteriorly and
shallow in transverse dimension. The frontal lappet of E.
novomexicanus extends farther laterally and has a reduced
contribution to the orbital margin (S. P. Modesto and R. R.
Reisz, personal observation). In E. boanerges this apparent
trend is continued and the frontal contribution to the orbital
margin is greatly reduced (personal observation).
In addition to the Shallow lateral lappet, the frontal of the
Garnett edaphosaur is primitive in featuring a relatively long
and broad anterior process, approximately half the length of the
frontal at the midline. The frontal of Ianth.asaurus is also
primitive in featuring weak yet discernible sculpturing. The
long anterior process, the transversely shon lateral lappet, and
weak sculpturing give the frontal an appearance similar to that
observed in Haptodus, the most primitive known sphenacodont.
The parietal of the Garnett edaphosaur resembles the parietal
of Haptodus and Ophiacodon more closely than that of Eda­
phosaurus. The lateral edge of the parietal is convex in dorsal
aspect and the posterior process is directed posteriorly, unlike
the condition seen in Edaphosaurus, where dorsal expansion of
the temporal fenestra embayed the parietal laterally, and the
posterior process of the parietal became posterolaterally direc­
ted following lateral expansion of the posterior cheek. In
addition, the parietals form a posterior median wedge in
lanthasaurus probably because of the relatively large size and
posterior location of the pineal foramen. Although the parietal
of lanthasallrus is similar to that of Haplodus (Currie 1977,
1979) in the respective convexity and concavity of its lateral and
posterior edges. the parietal of the Garnett edaphosauris distinct
in that it contacts thc squamosal lateral to the supratemporal.
The postorbital of lanthasaurus bears a small posterodorsal
process which is excluded from contacting the supratemporal by
the parietal. A short posterior process is also present in both
E. novomeXiCQflu..~ and F. hoaneTges (p::;rsonal observation), and
as in lanthasaurus, there is no evidence indicating contact with
the supratemporal posteriorly. A postorbital-supratemporal
cuntact may have indeed strengthened the nnion between the
cheek and skull table in a short-cheeked animal such as
Procolophon (Carroll and Lindsay 1985), but the absence of
such a contact in edaphosaurs implies that a long posterior
process of the postorbital did not necessarily develop in
associatiun with the development of a lateral temporal opening
as suggested by Carroll (1986).
Other aspects of the skull table imply a primitive morpholo­
gical grade for lanthasaurus. The postfromal of EdaphosauTus
has a broad anterior expansion, probably resulting from the
lateral expansion of the supraorbital shelf and possibly also
from the dorsal expansion of the temporal fenestra. Reisz and
Berman (1986) interpreted the postfrontal of lanthasaurus as
having a large anterior expansion as in Edaphosaurus. Unfor­
tunately, the orientation of the isolated postfrontal in ROM
29942 was misidentified by Reisz and Bennan (1986). ROM
37751 reveals that the posterior portion of the postfrontal is
more transversely broad than the anterior portion. This pattern is
also observed for the postfrontal in Haptodus, Archaeothyris,
and Eothyris. In relative size and configuration, the temporal
fenestra of !anthasaurus (Fig 4a) resembles closely that of
Haptodus. The size of the temporal fenestra relative to the cheek
in lanthasaurus is not significantly enlarged when compared
with the same opening in Edaphosaurus. The primitive con­
figuration of the postfronral, the short lateral lappet of the
frontal, the small size of the temporal opening, and the trans­
versely broad parietals necessitate the revised familial diagnosis
given in the Systematic Paleontology section. Until materials
presently unknown for lanthasaurus are available and proVIde
information that may expand diagnosis offered here, the family
must be defined only by the morphology of the distinctive
presacral neural spines.
The presence of a dorsal sail in the sma)] pelycosaur !antha­
saurUJ requires a reconsideration of the possible functions of the
distinctive edaphosaur dorsal sail. General concensuS holds that
the pelycosaur sail served for heat exchange rather than for
protection or sexual display. This hypothesis is supported by a
correlation between body volume and sail area in sphenacodon­
tids (Romer 1948). If the dorsal sail served as a thermoregUla­
tory device in edaphosaurs, one would expect a correlation
between sail area and body volume similar to that seen in
sphenacodonts. The reverse is true for Edaphosaurus, in which
relative sail area decreases with increasing body volume (Romer
and Price 1940; Berman 1979). lanthasaurus is distinguished
among edaphosaurs in possessing the smallest sail relative to
body size, a characteristic Reisz and Bennan (1986) considered
primitive for the family. In addition, the heat-exchange theory
does not account for the presence of the distinctive lateral
tubercles of the neural spines. The pattern of the crossbars
suggests that the sail may have also served for individual
recognition. This hypothesis is supported by the observation
that although the crossbars are distinctly set both vertically on
the same neural spine and horizontally across the series, the
pattern observed for ROM 37751 differs from that for KUVP
69035. Lastly, the edaphosaur sail may have hau a uefensive
purpose or served to make the animal appear larger and more
formidable. This is supported by four observations: (1) the
anterior and posterior neural spInes sweep over the neck and
hips, respectively; (2) the sail bears an imposing array oflateral
projections; (3) the lateral tnbercles typically reach their
greatest length proximally, projectlOg over the epaXIal muscn­
lature; and (4) the cervical and anteriol dorsal presacf1l1 neural
spines arc anteroposteriorly expanded in the most advanced
members of the family. To what degree the above hypotheses
accurately acconnt for the function of the edaphosaurid sail
remains highly speculative.
The elongate dorsal process on the axial neural spine
represents the second autapomorphy known for lanthasaurus.
This process increases the forward extent of the dorsal sail in the
Garnett edaphosaur (Fig. 5). The axial neural spine in lantha­
saurus apparently compensates for the cervical spines that are
less strongly swept forward than in other edaphosaurs. The
dorsal process of the axial neural spine probably had minimal
interference in the flexibility of the neck, outside that allowable
by the presence of the remaining cervical spines, yet may have
provided additional protection from predators in the anterior
region of the neck. In addition, the rugose ridge formed
posterior to the pineal opening on the parietals and the close
proximity of the axial neural spine to the skull surface suggest
that the dorsal sail may have extended to the dorsal surface of the
skull, to the posterior edge of the pineal opeuing.
The significant proportion of multiple tubercles and ventral
webbing on the basal tubercles of ROM 37751 and the
observation that the anterior presacral neural spines are 25%
longer than those of the holotypic sail raise the possibility that
lanthasaurus may have been sexually dimorphic. These differ­
ences, however, are not great enough to merit a separate specific
diagnosis. Multiple tubercles are not unique to lanthasaurus.
Carnegie Museum specimen CM 8540, attributed to Edapho­
saurus cruciger by Romer (1952). features a lateral tubercle that
clearly bears a small accessory process (see also Berman 1979).
However, no other cases of multiple tubercles of this nature
have been reported for Edaphosaurus.
Available dentition suggests that lanthasaurus fed primarily
on soft-bodied insects (Reisz and Bennan 1986). The sharp,
recurved teeth are not capable of shredding plant material as
leaf-shaped teeth are; the later are common to herbivorous
reptiles lacking massive, crushing tooth batteries. The skull and
mandible appear too lightly built to suggest a small camivore
role for !anthasaurus (Reisz and Bennan 1986). The adaptation
to insectivory in a small, primitive member within a group
comprising mostly large herbivores is strong testimony that
herbivory had evolved within the Edaphosauridae.
Current evidence accordingly suggests that the edaphosaurs
evolved first as small insectivores, bearing the characte11stic
cross-barred sail that distinguished them from primitive sphena­
codonts with which they shared a recent common ancestor.
The phylogenetic progression seen in the edaphosaurs is the
evolution of adaptations associated with herbivory. These
adaptations include modified dentition, the development of a
barrel-shaped trunk (containing an enlarged, possibly compart­
mentalized gut with endosymbiotic microorganisms), and the
evolution of large body size. These modifications are common
to extant herbivorous lizards and are not shared with carni­
843
MODESTO AND REISZ
vorous lizards (Zimmennan and Tracy 1989). The increase in
body size also may have been driven by intraspecific competi­
tion. On the other hand, the trend towards large size may have
instead been a response to a similar increase in body size of
their main predators, the sphenacodonts. This hypothesis is
supported hy the presence of a primitive sphenacodont at
Garnett. slightly larger than lanthasaurus and well advanced
towards a predaceous lifestyle (Reisz and Berman 1986).
Conversely, the trend towards large body size in sphenacodonts
may have been a response to the trend seen in euaphosaurs.
Consequently, it should be possible to identify edaphosaur
material, in the absence of dental indicators, as belonging to
either an insectivore or a herbivore. The postcrania of insec­
tivorous edaphosaurs may be identitled by small body size,
elongate trunk:, trunk ribs that are curved mainly proximally and
bear well-developed tubercula, and cervical centra of slightly
greater lengths than dorsal centra. Herbivorous edaphosaurs
may be identified in the absence of cranial material by large
size, cervical centra with lengths less than those of dorsal
vertebrae, and strongly curved trunk ribs lacking proper
tubercles. The strong curvature of the ribs implies the presence
of an enlarged, possibly compartmentalized digestive tract
bearing cellulytic microflora.
Four edaphosaur species are represented by fragmentary
postcranial material. Because the family is no longer monogen­
eric, possession of long neural spines with lateral tubercles
cannot be used to assign material to the genus Edaphosaurus. An
edaphosaur of medium size from the Lower Permian of West
Virginia, of which no cranial material is known, has been
eonfidently identified as Edaphosaurus colohistion on the basis
of strongly curved ribs and neural spine morphology character­
istic of the better known Texan edaphosaurs (Berman 1979). A
spine fragment from Pennsylvania, identified by Case (1908) as
Edaphosaurus raymondi, possesses no diagnostic characters
and is in all likelihood a nomen vanum. A partial vertebra
from Czechoslovakia, assigned to Edaphosaurus mirabilis.
may belong to the genus Iamhasaurus, for it is also Late
Pennsylvanian in age and of small body size. A partial skeleton
of the Early Pennian Edaphosaurus credneri, from the German
Democratic Republic is diminutive in size and features few
tubercles on the posterior presacral neural spines. It may also
belong to Ianthasaurus because of its small size aud the rarity
of tubercles on the posterior neural spines. Edaphosaurus
raymondi, E. mirabilis, and E. credneri need to be restudied.
If these species are reassigned to Iamhasaurus the geographic
and temporal ranges of this genus will be greatly increased,
perhaps equal to those of Haptodus.
The information provided' by Ianthasaurus supports the
hypothesis that the poorly known pelycosaur Lupeosaurus is an
edaphosaur. Recently, Sumida (1989) described new material
of this unusual pelycosaur. He noted that Lupeosaurus shares
several derived features with Ianthasaurus anu Edaphosaurus
yet remains distinct from the latter two genera in the absence of
lateral tUbercles of the neural spines and the possession of an
exaggerated, enlarged clavicular plate and a flared dorsal
scapular blade. Because of the absence of cranial materials,
Sumida (1989) tentatively referred the genus to the Edaphosau­
ridae, recognizing that the lack of lateral tubercles is possibly
a primitive condition for cdaphosaurs that may not necessarily
bar the inclusion of Lupeosaurus in the Edaphosauridae.
However, ROM 37751 reveals that no derived cranial charac­
Ters lITe known for edaphosanrs and suggeSTS That if Lupeo­
saurus cranial materials were to become aVU11able they would
be informative only at the subfamilial level. Nevertheless.
confident assignment of Lupeosaurus to the family must await a
thorough phylogenetic review of the Edaphosauridae.
lanthasaurus plays an important role in our understanding of
the adaptation to herbivory in early amniotes. The Gamett
edaphosaur possesses features that support the hypotheSIS that
herbivory had developeu withiu the family Edaphosauridae.
The known material of Ianchasaurus exhibits a number of
derived characters that suppon the hypothesis of an edapho­
saur~sphenacodont clade. This is further supported by the
observation tbat the skull-roof clements of this primitive eda­
phosaur ostensibly resemble those of the most primitive sphena­
codont, Haprodus garneuensis. The infonnation proVided by
the skull-roof elemeuts has shown also that the Edaphosauridae
can be defined only by the morphology of the presacral neural
spines. This study has stressed the need for the reexamination
of other early edaphosaur material. Future studies may reveal
lanthasaurus to be as geographically and temporally wide­
spread as its sphenacodont counterpart Haptodus.
Acknowledgments
We are indebted to Ms. Diane Scott, who provided much
needed technical assistance, initiated the preparation of the
specimen, reproduced the photograph in Fig. 1, and critically
read the manuscript. We thank: Mr. Michel Laurin for the
French translation of the abstract and his criticisms of the
manuscript. Thanks are also due Mr. David Dilkes, whose help
improved the final version of this paper.
BERMAN, D. S. 1979. Edaphosaurus (Reptilia, Pelycosauria) from the
Lower Pennian of northeastern United States, with description of a
new species. Annals of the Carnegie Museum, 48: 185-202.
BRINKMAN, D., and EBERTH, D. A. 1983. The interrelationships of
pelycosaurs. Breviora, 473: 1-35.
CARROLL, R. 1986. The skeletal anatomy and some aspects of the
physiology of primitive reptiles. In The ecology and biology of
mammal-like reptiles. Edited by P. N. Hotton, P. D. MacLean, J. J.
Roth, and E. C. Roth. Smithsonian Institution Press, Washington
and London, pp. 25-45.
CARROLL, R., and LINDSAY, W. 1985. Cranial anatomy of the
primitive reptile Procolophon. Canadian Journal of Earth Sciences,
22: 1571-1587.
CASE, E. C. 1908. Description of vertebrate fossils from the vicinity of
Pittsburgh, Pennsylvania. Annals of the Carnegie Museum, 4:
234-241.
CURRIE, P. J. 1977. A new haptodontine sphenacodont (Reptilia:
Pelycosauria) from the Upper Pennsylvanian of NOl1h America.
Journal of Paleontology, 51: 927-942.
- - - 1979. The osteology of haptodontine sphenacodontids (Rep­
tilia: Pelyeosauria). Palaeontographica, Abteinburg A, 163: 130-168.
DILKES, D. w. 1987. A new genus of pelycosaurian reptile from the
Upper Pennsylvanian of Garnett, Kansas. Journal of Vertebrate
Paleontology, Abstracts, 7: 15A.
KEMP, T. S. 1982. Mammal-like reptiles and the origin of mammals.
Academic Press, London.
REISZ, R. R. 1977. Petrolacosaurus, the oldest known diapsid reptile.
Science (Washington, DC), 1~6; 1091-1093.
- - 1980. The Pelycosauria; a review of phylogeneric relation­
ships. In The terrestrial envIronment and th~ origin of land verte­
brates. Edhed by A. L. Panchen. Systematics Association. Special
Symposium 15, pp. 553-592.
- - - 1981. A diapsJd reptile from the Pennsylvanian of Kansas.
University of Kansas Museum of Natural History. Special Publica­
tion 7, pp. 1-74.
844
CAN. 1. EARTH SCI VOL. 27, 1990
- - - 1986. Pelycosauria. In Handbuch del' PaHioherpelologie. ParI
17 A. Edited by P. WcJlnhofcr. Gustav Fischer Verlag, Stuttgart, pp.
1-102.
REISZ, R. R., and BERMAN, D. S. 1986. lamhasaul'Ils hardeslii II. sp.,
a primitive edaphosaur (Reptilia, Pe1ycosauria) from the Upper
Pennsylvanian Rock Lakc Shale near Garnell. Kansas. Canadian
Journal of Earth Sciences, 23: 77-91.
REISZ, R. R., HEATON, M. J., and PYNN B. R. 1982. Vertebrate
fauna of Lale Peumylvanian Rock Lake Shale ncar Garnett, Kansas:
J'elycosauria. Journal of Paleontology, 56: 741-750.
ROMER, A. S. 1948. Relative growth in pelycosaurian reptiles. In
Rohert Broom commemorative volume. Edited by A. L. Du ToiL
Royal Society of South Africa, Special Publication, pp. 45-55.
- - - 1952. Late Pennsylvanian vertebrates and Early Permian
vertebrates of the Pittsburgh - West Virginia region. Annals of the
Carnegie Museum, 33; 47-113.
ROMER, A. S., and PRICE, L. I. 1940. Review of the Pelycosauria.
Geological Society of America, Special Paper 28.
SUMfDA, S. S. 1989. New infonnation on the pectoral girdle and
vertebral column in Lupeosaurus (Reptilia, Pelycosauria). Canadian
Journal of Earth Sciences, 26: 1343-1349.
ZJMMERMAN, L. c.. and TRACY, C. R. 1989. Interaclions between the
environmenl and ectothermy and herbivory in reptiles. Physiological
Zoology, 62: 374-409.